Path planning method based on umbrella lizard optimization improved model

Abstract

The application discloses a path planning method based on a chameleon optimization improved model, relates to the intelligent obstacle avoidance technical field, and specifically relates to the path planning technical field based on the chameleon optimization improved model. The method is based on a dynamic and static cooperation method, can update an environment model in real time, and can plan a path which can avoid obstacles and is efficient by combining the chameleon optimization improved model. The method comprises the following steps: acquiring an environment information data set; constructing and improving the chameleon optimization model; in the development stage of the chameleon optimization model, adding an adaptive t-distribution variation strategy and a Levy path strategy in parallel between a conventional strategy for calculating a new position of a population individual and a population individual new position updating module; adding a Cauchy variation strategy between the population individual new position updating module and a population individual optimal position updating module; and planning a path through the trained chameleon optimization improved model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent obstacle avoidance, in particular to the technical field of path planning based on a frilled lizard optimization improved model. BACKGROUND

[0002] In the field of intelligent mobile devices, intelligent agents such as robots, unmanned vehicles and drones with tracking functions face multiple technical challenges in actual operation. As the application scenarios gradually expand to complex dynamic environments, such as urban traffic arteries, complex topography and densely populated places, the high complexity and rapid dynamic characteristics of environmental information make map data updating lag, obstacle avoidance path planning inaccurate, and other problems increasingly prominent. Due to the technical bottlenecks in real-time and accuracy of path obstacle avoidance planning, intelligent agents are difficult to respond to sudden obstacles in the travel path in a timely manner, resulting in frequent collision accidents.

[0003] In a complex dynamic environment, the spatial position and motion speed of obstacles are constantly changing, and traditional obstacle avoidance algorithms based on static environment assumptions have been difficult to meet actual needs. In recent years, the frilled lizard optimization algorithm (FLO) has provided a new solution for dynamic obstacle avoidance path planning. Although it can adapt to environmental dynamic changes to some extent, in actual application, the paths planned by it generally have serious redundancy. This redundancy not only increases the consumption of computing resources and prolongs the path planning time, but also leads to increased device operating energy consumption and reduced execution efficiency, resulting in significant waste of manpower, material resources and time costs, and restricting the application efficiency of intelligent agents in complex scenarios. SUMMARY

[0004] In view of the problems that the existing technology cannot update environmental information in a timely manner and the obstacle avoidance path has serious redundancy, the purpose of the present application is to propose a path planning method based on a frilled lizard optimization improved model. The method described in the present application is based on a dynamic and static cooperative method, which updates the environmental model in real time, and then combines the frilled lizard optimization improved model to plan a path that not only avoids obstacles but is also efficient.

[0005] The method comprises the following steps:

[0006] S1, obtaining an environmental information dataset;

[0007] S2, constructing and improving a frilled lizard optimization model:

[0008] S21, in the development stage of the frilled lizard optimization model, parallelly adding adaptive distribution mutation strategy and Levy path strategy between the conventional strategy for calculating the new position of the population individual and the population individual new position updating module;

[0009] setting an update strategy threshold , interval randomly selected in the interval

[0010] When , an adaptive distribution mutation strategy is executed;

[0011] Otherwise, a Levy path strategy is executed;

[0012] S22, in the development stage of the chameleon optimization model, a Cauchy mutation strategy is added between the population individual new position updating module and the population individual optimal position updating module;

[0013] S3, iteratively train the chameleon optimization improved model ;

[0014] S4, plan a path through the trained chameleon optimization improved model;

[0015] The path planning through the trained chameleon optimization improved model is specifically:

[0016] S41, determine whether the distance between the sudden obstacle and the current position is less than the safety distance, if less than the safety distance, execute step S42, otherwise, keep the original path;

[0017] S42, update the environment information dataset to trigger the trained improved chameleon optimization model to plan a path.

[0018] Further, the environment information dataset represents the dimensional path model of the environment to be planned.

[0019] Further, the adaptive distribution mutation strategy is specifically: , wherein represents the position of the first chameleon in the adaptive distribution mutation strategy in the development stage, represents the position of the first chameleon in the conventional strategy for calculating the new position of the population individual, represents the distribution with the iteration number, . .

[0020] Further, the Levy path strategy is specifically: , wherein represents the position of the first chameleon in the Levy path strategy in the development stage, represents the position of the first chameleon in the conventional strategy for calculating the new position of the population individual, represents the position offset coefficient, represents the Levy path mechanism,​​ represents the random path length.

[0021] Further, the population individual new position updating module is specifically:

[0022] wherein, represents the evaluation function value of , and respectively represent the evaluation function values of and .

[0023] Further, the Cauchy mutation strategy is specifically: wherein, represents the position of the i-th chameleon determined by the population individual new position updating module, represents the position of the i-th chameleon, represents the Cauchy distribution function, represents the position of the i-th chameleon in the Cauchy mutation strategy in the development stage.

[0024] Further, the population individual optimal position updating module is specifically:

[0025] wherein, represents the evaluation function value of , represents the evaluation function value of , represents the position of the i-th chameleon determined by the population individual optimal position updating module.

[0026] The method has the following beneficial effects:

[0027] (1) The method is based on a dynamic and static combined method, and the environment model (obstacle information) is updated in time to provide real-time and reliable environment information for path planning, and to ensure that the new path can timely avoid moving or newly added obstacles.

[0028] (2) The method can solve the problems of slow convergence speed and easy falling into local optimum of the basic chameleon optimization algorithm by introducing the Levy path strategy in the development stage of the chameleon optimization algorithm, and improve the global search ability.

[0029] (3) The method introduces an adaptive distribution mutation strategy in the development stage of the chameleon optimization algorithm, and performs ​​The distribution disturbance variation does not change the original umbrella lizard optimization algorithm updating principle formula, has good global development capability in the early iteration stage, has good local exploration capability in the later iteration stage, and thus the convergence speed of the umbrella lizard optimization algorithm is accelerated.

[0030] (4) The method introduces the Cauchy variation strategy in the development stage of the umbrella lizard optimization algorithm, disturbs the current optimal individual, ensures that the umbrella lizard optimization improved model can smoothly jump out of the local extreme value area, and thus the influence of the local optimum on the optimization ability of the umbrella lizard optimization improved model is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 The path planning flowchart of the application is shown in the figure;

[0032] Figure 2 The flowchart of the umbrella lizard optimization improved model of the application is shown in the figure;

[0033] Figure 3 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0034] The obstacle avoidance path schematic diagram of the application is shown in the figure; Figure 4 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0035] Figure 5 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0036] Figure 6 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0037] The obstacle avoidance path schematic diagram of the application is shown in the figure; Figure 7 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0038] Figure 8 The obstacle avoidance path schematic diagram of the application is shown in the figure;

[0039] Figure 9 The three-dimensional schematic diagram of path planning algorithm comparison test results in the first quantity obstacle map of the application is shown in the figure;

[0040] Figure 10 The top view (Top view) diagram of path planning algorithm comparison test results in the first quantity obstacle map of the application is shown in the figure, wherein, origin represents the starting point, and destination represents the end point;

[0041] Figure 11 The three-dimensional schematic diagram of path planning algorithm comparison test results in the second quantity obstacle map of the application is shown in the figure;​​​​

[0042] Figure 12 This is a top view of the results of the path planning algorithm comparison test in the second type of obstacle map of the present invention, where origin represents the starting point and destination represents the end point;

[0043] Figure 13 This is a three-dimensional schematic diagram of the comparative test results of the path planning algorithms in the third type of quantitative obstacle map described in the present invention;

[0044] Figure 14 This is a top view of the results of the path planning algorithm comparison test in the third type of quantitative obstacle map described in the present invention, where origin represents the starting point and destination represents the end point;

[0045] Figure 15 This is a schematic diagram of the path lengths planned by each model in the comparative test of the three maps (environments) with different numbers of obstacles described in the present invention. DETAILED DESCRIPTION

[0046] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0047] Example 1

[0048] This embodiment provides a path planning method based on the optimized improved model of the umbrella lizard. In this embodiment, the intelligent mobile device is an unmanned vehicle, such as Figure 1 As shown, the method includes the following steps:

[0049] Step 1: When the unmanned vehicle detects a sudden static obstacle, it immediately calculates the current path segment vector and the projection of the static obstacle onto the current path. Based on the above calculation results, it determines whether the static obstacle is inside or near the current path (line segment). If the static obstacle is not on the current path, it skips the current path segment and jumps to the next path segment.

[0050] Step 2: Once the static obstacle is confirmed to be on the current path, calculate the shortest distance from the static obstacle to the current path segment. ,like If the distance is greater than or equal to the safety distance threshold, the original path is maintained, but the area should be continuously monitored to prevent the future path from expanding into the area or static obstacles from moving onto the current path;

[0051] Step 3: If less than the safety distance threshold, immediately mark its position as impassable and update the environment information dataset (environment model) representing the environment in which the path is to be planned the path model;

[0052] trigger the chameleon optimization improvement model to re-plan the path based on the updated environment information dataset;

[0053] Step four, determine whether the shortest distance of the new path to the static obstacle is less than the safety distance threshold. If it is, trigger the chameleon optimization improvement model to re-plan the path. Otherwise, synchronize the new path to the environment information dataset.

[0054] Embodiment 2,

[0055] This embodiment is a further limitation of Embodiment 1.

[0056] When facing a single or multiple moving obstacles, the unmanned vehicle needs to predict the position of the obstacles in the future time based on the known speed and direction information. This process includes calculating the target distance, i.e. the distance the unmanned vehicle should walk in a given time, and accumulating the path distance to determine which segment of the path the unmanned vehicle should be located on, and finally calculating the specific position of the unmanned vehicle on this segment of the path. In this way, the process of the unmanned vehicle moving along the path can be effectively simulated. A time window is set and a small time step is adopted to gradually check the relative position between the unmanned vehicle and each moving obstacle. If the prediction shows that the distance between the two is less than the preset safety threshold, it is considered that there is a potential collision risk. At this time, the predicted collision point needs to be immediately marked as impassable and the environment information dataset is updated.

[0057] Based on the updated environment information dataset, a new safe path is planned using the chameleon optimization improvement model. This new path should avoid all areas where potential collisions may occur in the prediction, while minimizing the additional distance caused by path adjustment. It is worth noting that if the newly planned path still has potential collision risks, it needs to be adjusted again until a completely safe path is found. In addition, in order to ensure that the path of the unmanned vehicle is always safe, the above collision detection and path adjustment process is repeated every fixed time interval. Once a new dynamic obstacle or static obstacle is detected, the path re-planning process is triggered immediately, rather than waiting for the next periodic check. This continuous monitoring and immediate response mechanism is crucial for dealing with complex and changing environments, as it ensures that the unmanned vehicle can always move along the optimal and safest path.

[0058] Embodiment 3,

[0059] This embodiment is a further limitation of Embodiments 1 and 2.

[0060] Frilled Lizard Optimization (FLO) is proposed by Ibraheem Abu Falahah et al in 2024, which simulates the unique hunting behavior of frilled lizards in natural habitats. Therefore, the algorithm consists of three stages, namely population initialization, exploration stage and development stage.

[0061] The method improves the frilled lizard optimization model.

[0062] As shown in Figure 2 , the working process of the improved frilled lizard optimization model is as follows:

[0063] I. Population initialization

[0064] Initialize the frilled lizard population, and randomly initialize the population members in the search space. The population matrix can be represented by formula (1), wherein, represents the th dimension, represents the total number of search map dimensions, represents the population of frilled lizards, :

[0065] (1)

[0066] The position of the th frilled lizard can be represented by formula (2), wherein, represents the position of the th frilled lizard in the th dimension, represents a number randomly selected from the interval , and represent the lower and upper bounds of random walk in the th dimension, respectively: (2) The improved frilled lizard optimization model obtains the positions of the starting point and the ending point, sets the initial parameters of the improved frilled lizard optimization model, and sets

[0067] and , ,

[0068] represents the th iteration, represents the total number of iterations, when , the optimal path from the starting point to the ending point has been generated, and the iteration is ended, , represents the th frilled lizard, ​​​​​​​Represents the total number of frilled lizards (population members).

[0069] 2. Exploration Phase

[0070] One of the most iconic natural behaviors of frilled lizards is their hunting strategy. Frilled lizards are sit-and-wait predators that attack upon sighting their prey. Simulating the frilled lizards' movement toward their prey results in extensive changes in the positions of population members within the problem-solving space, thereby enhancing the algorithm's global search capabilities.

[0071] In the exploration phase, the positions of the individuals in the population are updated according to the hunting strategy of the frilled lizard. The method of the present invention takes the position of the population member with the best evaluation function value as the prey position, i.e. The new location of the frilled lizard can have multiple more favorable locations (prey). The new position of the frilled lizard can be expressed by formula (3), where Indicates the The new location of the frilled lizard, Indicates the The evaluation function value of the current position of the frilled lizard, Indicates the Only prey, Indicates the The location of the prey of the frilled lizard, express The evaluation function value of :

[0072] (3)

[0073] Evaluation function value representation: initial position Through the Fringed lizard Arrive at the target location The distance can be expressed by formula (4): Indicates the The position of the frilled lizard in the first dimension, Indicates the Fringed lizard in the The location of the dimension, Indicates the The frilled lizard Dimensional location, Indicates the The frilled lizard Dimension location:

[0074] (4)

[0075] Assume that the frilled lizard randomly selects one of the candidate prey and attacks it. Based on the modeling of the frilled lizard's movement toward the selected prey, the new position of each frilled lizard individual in the population during the exploration phase is calculated using formula (5), where Indicates the The new position of the frilled lizard during the exploration phase, Represents from the set Random numbers picked from:

[0076] (5)

[0077] judge The evaluation function value of Is it better than The evaluation function value of , and use formula (6) to calculate the The optimal position of a frilled lizard during the exploration phase :

[0078] (6)

[0079] 3. Development Phase

[0080] After feeding, frilled lizards retreat to the treetops near their location. By simulating the lizard's movement to the treetops, the positions of individual lizards in the solution space undergo slight changes, thereby improving the algorithm's ability to utilize local search. During the development phase, the positions of individual lizards in the solution space are updated based on the strategy of retreating to the treetops after feeding.

[0081] Based on the modeling of the behavior of frilled lizards moving to the top of nearby trees, the general strategy for calculating the new position of individuals in the population is shown in formula (7), where represents the conventional strategy for calculating the new position of individuals in the population. The location of the frilled lizard, and They represent the upper and lower bounds of the random walk respectively:

[0082] (7)

[0083] The basic Frilled Lizard Optimization (FLO) algorithm is an intelligent algorithm that simulates the unique hunting behavior of frilled lizards in their natural habitat. This algorithm has the advantages of strong global search capabilities and fast convergence, but it also has some shortcomings:

[0084] 1) Slow convergence: In some complex problems, the Flying Lizard Optimization Algorithm may require a large number of iterations to reach a satisfactory solution, which will affect the efficiency of the algorithm.

[0085] 2) Optimization accuracy problem: The basic umbrella lizard optimization algorithm may have insufficient accuracy in finding the optimal solution, especially in high-dimensional and complex optimization problems, which may lead to the solution not being globally optimal.

[0086] 3) Prone to local optimum: Although the umbrella lizard optimization algorithm has strong global search capability, in some cases, the algorithm may still fall into a local optimal solution, especially in search spaces with a large number of local optima.

[0087] 4) Parameter adjustment sensitivity: The performance of the basic umbrella lizard optimization algorithm depends on the setting of the initial parameters to some extent, and improper parameter selection may affect the search effect of the algorithm and the quality of the final solution.

[0088] 5) Adaptability problem: The algorithm may need to be adjusted and optimized for different optimization problems to adapt to the characteristics and needs of specific problems, which may increase the complexity of the algorithm application

[0089] To solve the above problems, the method described in the present application increases adaptive distribution mutation strategy and Levy path strategy between the conventional strategy for calculating the new position of the population individual (formula (7)) and the population individual new position update module (formula (13)).

[0090] adaptive distribution mutation strategy is as follows:

[0091] (8)

[0092] wherein, represents the development stage in the adaptive distribution mutation strategy, the position of the umbrella lizard, has been mutated, indicating that the distribution with the number of iterations is considered as a parameter degree of freedom. Based on this definition, a distribution random disturbance term is introduced to fully utilize the information disturbance of the current population. The initial number of iterations is moderate, the distribution mutation is similar to the Cauchy distribution mutation, with strong global search performance; in the later stage, it is similar to the Gaussian distribution mutation, with excellent local development potential. In the middle stage of the algorithm, the distribution mutation is between the Cauchy mutation and the Gaussian mutation. The distribution mutation operator has the advantages of Gaussian and Cauchy operators, and optimizes the global exploration and local exploration performance of the algorithm.

[0093] Levy flight strategy is a random walk process used to simulate animal foraging. Many researchers apply this strategy to solve random search problems to optimize algorithm performance and obtain better models. The advantage of Levy flight strategy is its search diversity and long jump characteristics, which can effectively avoid local optimal solution and explore the entire search space. At the same time, it has better convergence performance and robustness, and is suitable for different types of optimization problems.

[0094] Levy flight strategy is specifically:

[0095] (9)

[0096] wherein, represents the position of the first chamaeleo in the development stage in the Levy flight strategy, represents the position offset coefficient, (10), represents a randomly selected number in the interval represents a randomly selected number in the interval represents the Levy flight mechanism, represents the random path length, (11), , both represent a random number in the range represents a constant, and in the present embodiment, the value of is 1.5, represents a parameter used for represents the random path length,

[0097] (12)

[0098] The update strategy threshold is set to , represents a randomly selected number in the interval

[0099] When , the adaptive distribution mutation strategy is executed;

[0100] otherwise, the Levy flight strategy is executed;

[0101] The position of the first chamaeleo is updated after the adaptive distribution mutation strategy or the Levy flight strategy, and then the population individual new position update module (formula (13)) is used to obtain the position of the first chamaeleo determined by the population individual new position update module:

[0102] ​​​​​ (13)

[0103] In the basic frilled lizard optimization algorithm, all frilled lizard individuals will gather towards the optimal individual when searching for the best. However, in the later stages of the algorithm, the diversity of the population will decrease, which will cause the algorithm to easily fall into a local optimal state. The Cauchy mutation strategy is introduced to perturb the current optimal individual to ensure that the algorithm can successfully jump out of the local extreme value area, thereby reducing the impact of the local optimality on the algorithm's optimization ability. The Cauchy mutation outlier is derived from the Cauchy distribution, and its function expression is shown in formula (14), where, express Input, Represents the Cauchy distribution probability function:

[0104] (14)

[0105] As shown in formula (15), the Cauchy mutation operator is introduced into the improved optimization algorithm for frilled lizards, and its perturbation ability is fully utilized to adjust the objective function value of the optimal frilled lizard individual:

[0106] (15)

[0107] in, Represents the first position determined by the population individual new position update module The location of the frilled lizard, represents the Cauchy distribution function, Indicates the development stage in the Cauchy mutation strategy The location of the frilled lizard.

[0108] Will Input the optimal position update module of the population individual (Formula (16)) to obtain the development stage The optimal position of the step, if , then save the current best candidate solution ( The best positions of the first step are spliced ​​together, which is the current best candidate solution). Otherwise, , No. Only the frilled lizard performs optimization during the exploration and exploitation phases.

[0109] The optimal position update module of the population individual is shown in formula (16):

[0110] (16)

[0111] in, express The evaluation function value of express The evaluation function value of represents the first position determined by the population individual optimal position update module Position of chamaeleo.

[0112] Determine whether the iteration is completed, i.e., if the iteration is completed, output the current best candidate solution (xk, f(xk)) representing the minimum path cost and the optimal path planning.

[0113] Otherwise, if the first optimization is executed.

[0114] Embodiment 4,

[0115] This embodiment is a further limitation of embodiments 1 to 3.

[0116] This embodiment sets the simulation experiment scene to include fixed obstacles, sudden obstacles and moving obstacles. The fixed obstacles, such as walls and buildings, are marked with black crosses in the simulation map to clearly identify the impassable areas for the robot. The sudden obstacles are randomly generated and appear in different positions on the map in the form of fixed red triangles, which increases the uncertainty of path planning and the requirement for dynamic response. The moving obstacles have dynamic characteristics and can move according to the preset trajectory in the simulation environment. These obstacles are usually represented by other robots or dynamic objects and are represented in the form of moving red rectangles on the simulation map to test the obstacle avoidance ability of the robot when facing dynamic obstacles. The forbidden area is a certain area that is set as a forbidden area due to safety or other reasons. In the simulation environment, these areas are intuitively marked with red crosses.

[0117] In this embodiment, the intelligent mobile device is a robot, which starts from the starting point S and aims at the terminal point T.

[0118] As shown in Figure 3 , at the initial moment, the robot uses the first optimal path calculated by the chamaeleo optimization improved algorithm and starts to travel along the path.

[0119] As shown in Figure 4 , at the moment, the robot faces an obstacle moving from the obstacle starting point to the obstacle terminal point in the direction, according to the set prediction time window analysis, if the current path is not changed, the robot will collide with the moving obstacle. Therefore, as shown in Figure 5 , at the moment, the area between the two red cross points is identified as a potential collision interval point and marked as a temporary forbidden area, the environment information dataset is updated and the path re-planning operation is executed to ensure that the robot can safely avoid the obstacle. ​​

[0120] like Figure 6 As shown, the robot When encountering a fixed sudden obstacle red triangle at any time, the environment information dataset is updated immediately and the path is replanned based on the new environment information to avoid the obstacle.

[0121] like Figure 7 As shown, the robot At this moment, I encountered another obstacle starting point. Towards the end of the obstacle Moving obstacles, according to the time window prediction, if the path is not adjusted, the robot will collide with the obstacle, e.g. Figure 8 As shown, in At this moment, the red cross area is marked as a temporary no-entry area, the environment model is updated and the path is replanned to avoid collision with moving obstacles, ensuring that the robot can reach the destination T smoothly.

[0122] arrive Throughout the entire process, the dynamic obstacle avoidance strategy of the method described in the present invention demonstrates its efficiency and reliability in dealing with complex and changing environments, and emphasizes its ability to respond quickly to emergencies and its high adaptability.

[0123] Example 5

[0124] This embodiment further limits Embodiments 1 to 4.

[0125] The method of the present invention can also be applied to planning the obstacle avoidance path of a UAV.

[0126] Example 6

[0127] This embodiment further limits Embodiments 1 to 5.

[0128] In order to verify the effectiveness of the improved frilled lizard optimization algorithm (IFLO), this example uses three maps with different numbers of obstacles for comparative experiments. The improved frilled lizard optimization algorithm (IFLO), the raccoon optimization algorithm (COA), the frilled lizard optimization algorithm (FLO), the sparrow search algorithm (SSA), the dung beetle optimization algorithm (DBO), and the pelican optimization algorithm (POA) respectively perform global path planning experiments in a three-dimensional environment model. The experimental results are shown in Figure 2. Figures 9 to 14 shown.

[0129] In this embodiment, the population size of COA, FLO, IFLO, SSA, DBO and POA is set to 30, and the maximum number of iterations is set to 500. The practicability of IFLO and its influence and performance on model performance are verified.

[0130] As shown in Figures 9 to 14 , IFLO and other models can achieve the conventional path planning from the starting point to the end point, avoiding obstacles and dangerous areas in the environment. However, although COA can complete the path planning of the unmanned aerial vehicle, the path planned by it has too many uneven phenomena, the path fluctuates greatly, and the overhead (two-dimensional) image shows that the distance to the dangerous area is too close. These problems show that the COA falls into local optimization in the path planning process, resulting in a high cost of the finally planned path and poor quality. In contrast, the path planned by IFLO is more stable, the fluctuation is moderate, the path is overall smooth, and a safe distance is maintained from the dangerous area.

[0131] The path lengths planned by each model in the comparative test of three different obstacle maps (environments) are shown in Figure 15 : the path planned by IFLO is the shortest in the three different obstacle maps, and the paths planned by COA and DBO can meet the requirements, but they are not as good as the planning trajectory of IFLO because of their longer distance and greater energy consumption. IFLO can obtain a better planning path than the other five algorithms in different complexity scenarios. In this embodiment, the performance of IFLO is detected by setting environments of different complexity (different obstacle maps). The test results prove that IFLO has the ability to plan paths in complex environments and has good performance and shorter execution time. This proves to a certain extent the effectiveness and superiority of the performance of IFLO.

[0132] The intelligent mobile device path planning technology is one of the core research contents in the field of intelligent mobile device application, which involves designing an optimal or feasible path for intelligent mobile devices from the starting point to the end point, while considering multiple conditions such as driving safety, efficiency, environmental factors and driving constraints. With the wide application of intelligent mobile devices in military reconnaissance, disaster rescue, environmental monitoring, agricultural plant protection, logistics distribution and other fields, the importance of path planning technology is increasingly prominent. It not only improves the autonomous driving ability of intelligent mobile devices and reduces manual intervention, but also enhances the ability of intelligent mobile devices to deal with complex environments and unexpected situations. In addition, optimized path planning can effectively reduce flight risks and improve the efficiency and quality of task execution, which has a profound impact on the healthy development of the intelligent mobile device industry. In global path planning, compared with traditional path planning algorithms, swarm intelligence optimization algorithms have wide application, strong scalability and great improvement significance. Therefore, the present application designs a path planning method based on the improved model of chameleon optimization algorithm, which is the best in the swarm intelligence optimization algorithm.

[0133] The present application aims at the problem that the intelligent mobile device will have poor optimization result and low search stability when planning a path, and proposes an improved chameleon optimization algorithm. In the chameleon pursuit and escape stage, the Levy flight strategy and adaptive t-distribution mutation strategy are used to effectively improve the accuracy of the solution and make the optimization effect better, and the local exploration ability in the later iteration is enhanced, thereby improving the convergence speed of the chameleon optimization algorithm. Finally, by introducing the Cauchy mutation strategy, the randomness of the improved chameleon optimization model is strengthened, which helps the improved chameleon optimization model to jump out of the local optimal solution and enhance the global search ability. On the basis of the above improved chameleon optimization model, it is applied to solve the intelligent mobile device path planning problem, and the experiment is carried out to verify the advantages of the model in solving the intelligent mobile device path planning problem. The flight path planned by IFLO is more stable, the fluctuation amplitude is moderate, the overall flight path is stable, and the flight path keeps a safe distance from the threat area. Compared with other five algorithms, the flight distance is shorter and the energy consumption is smaller, which shows that the convergence speed of IFLO is relatively fast and the stability is high, and it is verified that it is effective to apply IFLO to the global path planning problem of unmanned aerial vehicle. The above results show that the optimization ability and the ability to jump out of the local optimal ability of IFLO are improved compared with other models.

Claims

1. A path planning method based on the optimized model of the frilled lizard, characterized in that: The method comprises the following steps: S1. Obtain environmental information dataset; S2. Build and improve the lizard optimization model: S21. During the development phase of the optimization model for the frilled lizard, an adaptive strategy was added in parallel between the conventional strategy for calculating the new positions of individuals in the population and the module for updating the new positions of individuals in the population. Distribution mutation strategy and Levy path strategy; Setting the update policy threshold , Representation interval A randomly selected number from when When performing adaptive Distribution mutation strategy; Otherwise, the Levy path strategy is executed; S22. During the development phase of the optimization model for the frilled lizard, a Cauchy mutation strategy is added between the population individual new position update module and the population individual optimal position update module. S3. Iterative training of the lizard to optimize the improved model Second-rate; S4, optimize and improve the model planning path through the trained frilled lizard; The path planned by the trained optimization and improvement model of the frilled lizard is specifically as follows: S41, determine whether the distance between the sudden obstacle and the current position is less than the safety distance, if less than the safety distance, execute step S42, otherwise, maintain the original path; S42: Update the environmental information dataset to trigger the trained improved frilled lizard optimization model to plan a path.

2. The path planning method based on the optimized improved model of the frilled lizard according to claim 1 is characterized in that: The environmental information dataset represents the environment of the path to be planned. dimensional path model.

3. The path planning method based on the optimized improved model of the frilled lizard according to claim 2 is characterized in that: The adaptive The distribution mutation strategy is as follows: ,in, Indicates that the development phase is adaptive In the distribution mutation strategy The location of the frilled lizard, Represents a number of iterations distributed, represents the conventional strategy for calculating the new position of individuals in the population. The location of the frilled lizard, .

4. The path planning method based on the optimized improved model of the frilled lizard according to claim 3 is characterized in that: The Levy path strategy is specifically as follows: ,in, Indicates that the development stage is the first in the Levy path strategy The location of the frilled lizard, represents the position offset coefficient, Represents the Levy path mechanism, represents the random path length.

5. The path planning method based on the optimized improved model of the frilled lizard according to claim 4 is characterized in that: The population individual new position update module is specifically: ,in, express The evaluation function value of and Respectively and The evaluation function value of .

6. The path planning method based on the optimized improved model of the frilled lizard according to claim 5 is characterized in that: The Cauchy mutation strategy is specifically: ,in, Represents the first position determined by the population individual new position update module The location of the frilled lizard, represents the Cauchy distribution function, Indicates the development stage in the Cauchy mutation strategy The location of the frilled lizard.

7. The path planning method based on the optimized improved model of the frilled lizard according to claim 6 is characterized in that: The optimal position update module of the population individual is specifically: ,in, express The evaluation function value of express The evaluation function value of represents the first position determined by the population individual optimal position update module The location of the frilled lizard.

Images